**Analysis of the Overlapped Electrochemical Signals of Hydrochlorothiazide and Pyridoxine on the Ethylenediamine-Modified Glassy Carbon Electrode by Use of Chemometrics Methods**

**Yue Zhang 1,**†**, Yan Zhou 2,**†**, Shujun Chen 3, Yashi You 3, Ping Qiu 3,\* and Yongnian Ni <sup>3</sup>**


Academic Editors: Clinio Locatelli, Marcello Locatelli and Dora Melucci Received: 1 June 2019; Accepted: 9 July 2019; Published: 11 July 2019

**Abstract:** In this work, the electrochemical behavior of hydrochlorothiazide and pyridoxine on the ethylenediamine-modified glassy carbon electrode were investigated by differential pulse voltammetry. In pH 3.4 Britton-Robinson (B-R) buffer solution, both hydrochlorothiazide and pyridoxine had a pair of sensitive irreversible oxidation peaks, that overlapped in the 1.10 V to 1.20 V potential range. Under the optimum experimental conditions, the peak current was linearly related to hydrochlorothiazide and pyridoxine in the concentration range of 0.10–2.0 μg/mL and 0.02–0.40 μg/mL, respectively. Chemometrics methods, including classical least squares (CLS), principal component regression (PCR) and partial least squares (PLS), were introduced to resolve the overlapped signals and determine the two components in mixtures, which avoided the troublesome steps of separation and purification. Finally, the simultaneous determination of the two components in commercial pharmaceuticals was performed with satisfactory results.

**Keywords:** differential pulse voltammetry; hydrochlorothiazide; pyridoxine; chemometrics

#### **1. Introduction**

Hypertension is an independent disease characterized by blood pressure above the normal range, and its etiology is not yet clear. According to the World Health Organization (WHO), hypertension is defined as adult blood pressure exceeding 21.3/12.6 kPa (160/95 mmHg) [1,2]. As a common cardiovascular disease, the risk of hypertension is not only revealed in high blood pressure, but also in a variety of pathophysiological changes, including changes in cardiovascular structure and function, nephropathy, encephalopathy and retinopathy, causing coronary atherosclerosis, cerebrovascular sclerosis and ultimately life-threatening [3–5]. There are an estimated one billion cases of hypertension in the world, which is predicted to rise to over 1.5 billion by 2015 [6,7]. In China economic development has changed people's habits and diet, the elderly population has grown and the incidence of hypertension has risen to 27.8% [2,8]. Clinical studies have found that reasonable drug treatment can effectively control blood pressure and various symptoms [9,10]. Moreover, it can reduce the incidence of some related diseases, thereby improving people's quality of life [11].

Among hypertensive patients, it is found that a single antihypertensive drug cannot lower blood pressure to normal, so two or more antihypertensive drugs are necessary to make blood pressure normal [12–14]. Each of the drug components of compound medicines have synergistic effect, which not only enhances the antihypertensive effect, but also counteracts the adverse side effects [15–17]. Most

compound antihypertensives contain the diuretic-antihypertensive drug hydrochlorothiazide and a small amount of vitamin B6 (pyridoxine) to protect the heart and prolong the antihypertensive effect.

Hitherto, various analytical methods have been reported for the separation and determination of the components in compound preparations. It has been reported that HPLC can be used in combination with various detection techniques, such as electrochemical detection [18], mass spectrometry [19,20], desorption electrospray ionization [21], size-exclusion chromatography [22], chemiluminescence [23], or spectrophometry [24]. The electroanalytical chemistry method, with its advantages of higher sensitivity and selectivity, is preferable, and it has become a common method for drug analysis [25]. In electrochemistry, there are sereval ways to avoid overlapping signals: (1) selection of an appropriate pH and electrolyte; (2) analyte derivation; and (3) application of chemometrics. Among them, statistical chemometrics methods are environmentally friendly, and can avoid the troublesome steps of separation and purification. However, complex background currents often exist in voltammetric and polarographic analysis of drugs. The oxidation or reduction processes of electrodes are often irreversible, and their peak potentials will change with the change of drug concentration. This non-linear additive system will have a serious impact on the determination [26–28]. Moreover, it is extremely difficult to directly measure the content of a single component with serious overlap in a complex mixture. Therefore, the application of chemometrics in pharmaceutical electroanalytical chemistry has attracted extensive attention [29]. Classical least squares (CLS), principal component regression (PCR) and partial least squares (PLS) are commonly used [30].

In this work, the differential pulse voltammetry technique was used to investigate the electrochemical behavior of hydrochlorothiazide and pyridoxine on an ethylenediamine-modified glassy carbon electrode. The fabricated electrode has good stability and reproducibility [31,32], and the modified version has better electrochemical charateristics [33]. Their overlapping peaks were resolved by chemometrics, which avoided the tedious step of separation and purification, simplified the determination process and realized the simultaneous determination of two components in commercial pharmaceuticals.

#### **2. Results and Discussion**

#### *2.1. Voltammetry at a Modified Electrode*

Figure 1 shows the differential pulse voltammograms of 0.4 μg/mL hydrochlorothiazide and 0.2 μg/mL pyridoxine at the ethylenediamine-modified glassy carbon electrode (Figure 1a) and bare glassy carbon electrode (Figure 1b), respectively, in pH 3.4 B-R buffer.

**Figure 1.** The differential pulse voltammograms (DPV) of 0.4 μg/mL hydrochlorothiazide and 0.2 μg/mL at the ethylenediamine-modified glassy carbon electrode (**a**) and bare glassy carbon electrode (**b**) respectively in the pH 3.4 B-R buffer.

The voltammetric curves of each compound and their mixtures show the maximum peak potentials for hydrochlorothiazide and pyridoxine at 1.11V and 1.22V, respectively, as well as the heavily overlapped nature of the composite voltammograms of the mixtures. This indicated that the sensitivity of the two drugs on the modified electrode was stronger than that on the bare electrode, and the peak potential difference between the two drugs is larger, which was helpful for chemometric analysis.

Ethylenediamine modifies the surface of the glassy carbon electrode by covalent bonding, which is based on oxidation of amino groups [34,35]. Primary and secondary amines can form amine radical cations through the oxidation of the amino group in anhydrous ethanol or acetonitrile solution. The radical is further bonded on the surface of glassy carbon or carbon fibers to form C-H covalent bonds [36,37]. Therefore, the modification method is also called the "amine radical cation method".

It is generally believed that the modifying cation and the electrode are vertically oriented. Primary amines are easily grafted on the surface of electrode during the electrooxidation, while secondary amines oxidize to form a monolayer with low coverage, and tertiary amines cannot form a monolayer. Therefore, it can be concluded that the steric hindrance of the substituent affects the active site of radicals on the surface of the electrode during the amine oxidation process. The bonding reaction requires protons to be detached from the amine cation radical, so this reaction path is not feasible for tertiary amines [38]. Ethylenediamine is a typical aliphatic diamine, which exhibits characteristics of both primary and secondary amines. The chemical properties of ethylenediamine are related to the number of hydrogen atoms substituents on the functional group (NH2). In this experiment, it seems the special chemical structure coordinates on the surface of the electrode to form an ethylenediamine-modified electrode.

As for the repeatability of the modified electrode, the results showed that the relative standard deviation (n = 3) was less than 3.8%. Regarding stability, samples containing the drugs were stored in the refrigerator at 4 ◦C. After 15 days, the current was measured with a standard deviation of less than 3.4%. The results proved that the modified electrode was stable and the processes have good repeatability.

#### *2.2. Selection of Bu*ff*er Solution*

This experiment investigated the effects of various buffer solutions on the electrochemical response of drugs. The tartaric acid-sodium tartrate buffer, acetic acid-sodium acetate buffer, disodium hydrogen phosphate-citric acid buffer and Britton-Robinson buffer were investigated. It was found that the sensitivity and symmetry of the oxidation peaks of each component in the Britton-Robinson buffer were highest, so the B-R buffer was selected.

The effect of acidity on the voltammetric curves of the drugs was also investigated. The electrochemical voltammetric curves of hydrochlorothiazide and pyridoxine in a series of B-R buffer solutions at pH 1.98–11.98 were measured, respectively. It was found that the peak current of the drugs increased and then decreased with the increase of pH value, and the peak shape gradually broadened. The peak potential of the drugs moved negatively with the increase of pH and showed a linear relationship. The linear equations were expressed as below:

$$\text{Hydrochloricide: } E\_{\text{Pl}} = -0.0468 \, pH + 1.3202 \, (R = 0.999), \tag{1}$$

$$\text{Pyridoxine: } E\_{\text{P}} = -0.0515 \, pF + 1.2710 \, (R = 0.998), \tag{2}$$

In acidic B-R buffer (pH 3.4), the drugs had better peak shape and higher peak sensitivity. Thus, these results clearly indicated that the pH 3.4 B-R buffer solution should be selected as supporting electrolyte in this experiment.

#### *2.3. Cyclic Voltammetry to Study the Adsorptivity of the Electrode Reaction*

The effect of the scan rate and the square root of the scan rate on peak current by cyclic voltammetry for the different drugs is examined (Figure 2). The peak current of both components showed a linear

relationship with the change of scan rate, but had a upward bending curve with square root of scan rate, illustrating that the electrode process is controlled by the adsorption rate at that time.

**Figure 2.** Effect of scan rate (**a**) and square root of scan rate (**b**) on peak current of the drugs.

#### *2.4. Linearity Ranges and Limits of Detection for Drugs*

Under the optimal experimental conditions, the drug concentrations had linear relationships with the electrochemical signals. Figure 3 shows voltammetric curves for the determination of hydrochlorothiazide (a) and pyridoxine (b). There is a good linear relationship between the peak current and the drugs concentration in the range of 0.02–0.40 μg/mL for pyridoxine and 0.10–2.0 μg/mL for hydrochlorothiaxide. The regression parameters are summarized in Table 1, and the accuracy of the determination of these drugs has also been established by analysing the minimum concentration calibration graph for six known solutions, i.e., 0.02 μg/mL (see Table 1).

**Figure 3.** Voltammogram of hydrochlorothiazide (**a**) and pyridoxine (**b**) for different concentrations.



The relative standard deviation (R.S.D.) values obtained were 1.7% and 2.1% for pyridoxine and hydrochlorothiazide, respectively, and the detection limit values were 7.00 and 37.60 ng/mL for pyridoxine and hydrochlorothiazide, respectively. Therefore, it can be concluded that the proposed electrochemical analysis method for the determination of individual drugs is reliable.

#### *2.5. Prediction of Synthetic Mixtures of Drug Compounds*

Before the determination of unknown mixtures by chemometrics, a set of mathematical models should be developded to correct the concentration of each component in the mixtures. According to the four-level orthogonal array design represented by OA16(45), a set of standard samples was prepared, which indicated that a data set of 16 samples was required. The calibration set is based on the given concentration range for 0.02–1.0 μg/mL, as it is usually desirable to test the lower performance of the calibration rather than the higher concentrations [39]. The prediction power of the calibration model was then assessed using another set of samples consisting of 12 synthetic mixtures. In our work, CLS, PLS, and PCR models were established and their prediction errors were compared. CLS method is often called K matrix method, which is a commonly used multivariate correction method. This method is based on multiple linear regression and is frequently used for quantitative voltammetric analysis, which was a much common multivariate calibration method [40]. PCR and PLS, which are powerful multivariate statistical tools and are available as commercial software for laboratory computers [41], are based on factor analysis [42]. These methods have many of the full-soectrum advantages of the CLS method and have also been successfully applied for voltammetric analysis [43].

From the measured results in Table 2, the best prediction results could be obtained by the PLS method, while the CLS method was the worst. Because of the background current and the interaction between components in voltammetric analysis, the non-linear superposition of voltammograms would occur and the CLS method had some difficulties in analyzing such systems, while PCR and PLS are calibration methods based on factor analysis, which could better solve this kind of non-linear problem [44]. In the PCR and PLS methods, a certain number of factors must be selected to perform a matrix factorization, so that the response matrix could be reproduced within the experimental error range. The selection of factors has a great impact on the calculation results. Generally, when the number of factors was less than or slightly larger than the component number, the relative standard deviation of the calculation results was smaller [45]. Figure 4 shows the relationship between the relative error and the number of factors. In this way, when the PCR method was used, the number of factor was 4, the relative error was smaller and the relative error is the minimum when the number of factor was 4 by PLS method.


**Table 2.** Prediction results for pyridoxine(PX) and hydrochlorothiazide(HC) validation samples by different chemometrics methods (μg/mL).


**Table 2.** *Cont.*

<sup>1</sup> The factors of PLS is 4. <sup>2</sup> The factors of PCR is 4. <sup>3</sup> RPES (%) and RPET (%) are relative prediction errors for single and total components, respectively. <sup>4</sup> Recovery (%) <sup>=</sup> <sup>100</sup> × - (cij,pred − cij,added)/*n*, where n is the number of samples, cij is the concentration of the *j-*th component in the *i-*th sample.

**Figure 4.** The relationship of PREt and the number of factors used for PLS and PCR.

#### **3. Detection of the Drugs in Real Samples**

The proposed method was applied for the determination of the drugs in four commercial tablets, produced by two different pharmaceutical companies. In each case, the sugar coating was removed from ten tablets which were then ground into a powder and dissolved in doubly distilled water. After filtering three times the filtrate was collected and diluted to 50 mL with distilled water.

According to the described PLS procedure, the samples were analyzed with the results listed in Table 3. The values measured by the proposed method were consistent with the target values. The average relative error between two methods was 3.1%, indicating the method provided excellent precision in detecting the drugs. The recovery rate was between 92.5% and 106.2%, with RSD (n = 5) less than 2.9%. It could be seen that the simultaneous determination of hydrochlorothiazide and pyridoxine in compound medicines could be achieved by the proposed electrochemical method without labelling, and the results were satisfactory.


**Table 3.** Pyridoxine(PX) and hydrochlorothiazide(HC) in commercial tablets as determined by the proposed method (mg/tablet).

<sup>1</sup> Jinhua Yexing Pharmaceutical Co., Ltd., Shanxi. Lot code. 061108. <sup>2</sup> Dongzhitang Pharmaceutical Co., Ltd., Anhui. Lot code. 06101102. <sup>3</sup> Yinhe Pharmaceutical Factory, Jilin. Lot code. 20050801. <sup>4</sup> Fenhe Pharmaceutical Co., Ltd., Shanxi. Lot code. 0701031.

#### **4. Materials and Methods**

#### *4.1. Reagents and Apparatus*

Hydrochlorothiazide (0.1 mg/mL) and pyridoxine (0.02 mg/mL, ethanol) were obtained from Shanghai Biochemical Reagent Co. Ltd. (Shanghai, China). Ethylenediamine (>99%) was dissolved to 1 mol/L solution with anhydrous ethanol for use as the ethylenediamine-ethanol modifier. A Britton-Robinson (B-R) buffer solution of pH 3.4 was prepared by adding 15 mL 0.2 mol/L sodium hydroxide into 100 mL of a mixed acid containing 0.04 mol/mL each of orthophosphoric acid, acetic acid, and boric acid.

All chemicals used (Sinopharm Chemical Reagent Co., Ltd. Shanghai, China) were of analytical-reagent grade and all the solutions were prepared with doubly distilled water. The experiments were carried out at room temperature.

A CHI 660A electrochemical workstation (Shanghai CH Instruments Co., Shanghai China) equipped with a BAS C-1 cell stand was used for voltammetric measurements. The three electrode system consisted of a glassy carbon electrode as work electrode, Ag/AgCl reference electrode and platinum wire as auxiliary electrode. The pH measurements were performed on a SA720 meter (Thermo Orion, Waltham, MA, USA).

#### *4.2. Procedure*

#### 4.2.1. Modification of Glassy Carbon Electrode

Firstly the glassy carbon electrode was polished on 1200-grit Carbimet metallographic sandpaper, and to a mirror finish with 1.0 μm α-alumina powder, 0.3 μm α-alumina powder and 0.05 μm γ-alumina powder in turn. Then HNO3 (1:1), anhydrous ethanol and doubly distilled water were used to clean the electrode with ultrasonic irradiation, respectively. Finally, the activated electrode was dipped in 1 mol/L ethylenediamine-ethanol solution for 12 h and the ethylene- diamine-modified glassy carbon electrode was ready for use.

#### 4.2.2. Procedure

A suitable amount of standard solution of hydrochlorothiazide and pyridoxine or their mixed solution were prepared in the electrolytic cell, together with B-R buffer (2 mL pH 3.4). The solution was diluted to 10 mL with doubly distilled water and shaken evenly. Then the ethylenediamine-modified glassy carbon electrode was placed, the solution was thoroughly mixed by stirring for 15 s. After a 10 s static period, a differential pulse voltammetric scan was run from 0.9 to 1.4 V. The resulting voltammograms were sampled by a computer at 4 mV intervals. The experiment was completed at 20 ◦C room temperature.

#### **5. Conclusions**

In this work, the use of a modified electrode to improve the selectivity of voltammetric data in complex matrices is the issue, even if the separation is not complete and requires the use of chemometric tools. Furthermore, PCL and PCR seem to perform adequately for deconvolution of the voltammetric signals. Chemometrics methods can thus be used for the quantitative analysis of two components in commercial drugs with simple operation and satisfactory results.

**Author Contributions:** Y.Z. (Yue Zhang) substantially contributed the conception of the work; Y.Z. (Yan Zhou) drafted the work and contributed with Y.Z. (Yue Zhang) equally; S.C. analyzed the data; Y.Y. interpreted of data; P.Q. substantially contributed the design of the work and finally approved the published version; Y.N. agreed to be accountable for all aspects of the work.

**Funding:** This research was funded by the National Natural Science Foundation of China (21765015, 21808099) and the Jiangxi Province Food and Drug Administration Science Foundation (2016SP04), China.

**Conflicts of Interest:** The authors declare no conflict of interest. Sources of support is not in study. All the authors listed have approved the manuscipt. The decision had no such invedement.

#### **References**


**Sample Availability:** Samples of the hydrochlorothiazide and pyridoxine are available from the authors.

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Novel MIPs-Parabens based SPE Stationary Phases Characterization and Application**

**Angela Tartaglia 1, Abuzar Kabir 2,\*, Songul Ulusoy 3, Halil Ibrahim Ulusoy 4, Giuseppe Maria Merone 5, Fabio Savini 6, Cristian D'Ovidio 7, Ugo de Grazia 8, Serena Gabrielli 9, Fabio Maroni 1, Pantaleone Bruni 1, Fausto Croce 1, Dora Melucci 10, Kenneth G. Furton <sup>2</sup> and Marcello Locatelli 1,\***


Academic Editors: Victoria Samanidou and Alessandra Gentili Received: 24 July 2019; Accepted: 10 September 2019; Published: 13 September 2019

**Abstract:** In this work, the synthesis, characterization, and application of novel parabens imprinted polymers as highly selective solid-phase extraction (SPE) sorbents have been reported. The imprinted polymers were created using sol–gel molecular imprinting process. All the seven parabens were considered herein in order to check the phase selectivity. By means of a validated HPLC-photodiode array detector (PDA) method all seven parabens were resolved in a single chromatographic run of 25 min. These SPE sorbents, *in-house* packed in SPE empty cartridges, were first characterized in terms of extraction capability, breakthrough volume, retention volume, hold-up volume, number of theoretical plates, and retention factor. Finally, the device was applied to a real urine sample to check the method feasibility on a very complex matrix. The new paraben imprinted SPE sorbents, not yet present in the literature, potentially encourage the development of novel molecularly imprinted polymers (MIPs) to enhance the extraction efficiency, and consequently the overall analytical performances, when the trace quantification is required.

**Keywords:** MIPs; parabens; biological matrix; extraction procedure; HPLC-PDA; stationary phase characterization

#### **1. Introduction**

Parabens are alkyl esters of *p*-hydroxybenzoic acid, commonly used as antimicrobial agents and preservatives in food products, pharmaceutical preparations, and cosmetic products [1]. This family includes methyl paraben (MPB), ethyl paraben (EPB), iso-propyl paraben (iPPB), propyl paraben (PPB), iso-butyl paraben (iBPB), butyl paraben (BPB), and benzyl paraben (BzPB). Their molecular formula, molecular weight, and chemical structure are shown in Table 1. Low cost, broad spectrum of activity, and chemical and thermal stability explain their widespread application compared to other alternatives [2].


**Table 1.** General chemical–physical characteristics of parabens and their chemical structures.

Humans may be exposed to parabens through inhalation, dermal contact, and ingestion. Parabens have been considered low toxicity compounds; nevertheless, there is growing evidence of their implication as endocrine disruption compounds (EDCs), i.e., products with negative effects on endocrine system due to their interaction with production, release, transport, metabolism, or elimination of hormones. Many studies describe the toxic effects of parabens such as obesity, diabetes, breast cancer and problems in reproductive system [3].

According to the European Union (EU) and the United States Food and Drug Administration (US FDA), the concentration of parabens should not exceed 0.4% for a single paraben and 0.8% for a mixture of parabens in the cosmetics product; the concentration in drugs must be less than 0.1% and 0.3% for individual paraben and for parabens mixture, respectively [4].

During the last years, new methods have been developed for the determination of parabens in different matrices. Due to the low concentrations of these compounds, the sample preparation represents the critical step for the isolation and analysis of these target molecules. Usually, common extraction techniques were used for pre-concentrating parabens, such as solid-phase extraction (SPE) [5–8] and solid-phase microextraction (SPME) [9]. For SPE, different materials are available for extraction, preconcentration, and sample cleaning. Such sorbents used in solid-phase extraction can be divided into three main groups: silica sorbents, polymeric sorbents, and activated or graphitized carbon. Analytical parameters such as selectivity, affinity, and extraction capacity are closely associated with the sorbent used and, consequently, various SPE materials have been developed to replace the classic ones and to increase the extractive selectivity. An example is represented by molecularly imprinted polymers (MIPs) used as SPE sorbent [1].

MIPs are formed by fixing a molecule (called a "template") on the polymer; the molecule is extracted afterward, leaving complementary cavities behind [10]. Molecular imprinting is obtained by polymerization of monomers in the presence of template molecules (Figure 1). The imprinting template should be stable under polymerization reaction. The link between template and monomer must be strong to form the pocket but also weak enough to be able to remove the template at the end. Usually, the monomer and template are in 4:1 molar ratio. The most frequently used monomers are methacrylic acid and 4-vinylpyridine [11].

**Figure 1.** Synthesis of molecularly imprinted polymers (MIPs).

There are different methods used to perform molecular imprinting: covalent imprinting, non-covalent imprinting, and semi-covalent imprinting. In covalent imprinting, the template and monomer are linked by covalent bonds; after polymerization, the template is removed from the polymer through the cleavage of covalent bond. The template is replaced from target analytes that bind forming the same covalent links. In non-covalent imprinting, template and monomer are linked by non-covalent interactions such as hydrogen bonding, ionic interactions, π–π interactions, and van der Waals forces. After polymerization, the template is removed, and the target analytes bind the polymer via same non-covalent interactions. This technology is the most popular for the synthesis of MIP. In semi-covalent approach, the template is linked covalently to the monomer but is rebinding by non-covalent interactions [12].

Among several MIP synthesis approaches, sol–gel synthesis process for creating MIPs has drawn tremendous attention due to ruggedness of the polymers originated from hybrid inorganic–organic components, high thermal and solvent stability, and ability to maintain imprinted cavities even after multiple use as well as their characteristic high imprinting factor (IF) compared to organic MIP synthesis approach. The success of sol–gel based MIP synthesis primarily depends on the judicious selection of sol–gel functional precursor(s), networking precursor, and the reaction conditions including the catalysts. In the current study, 3-aminopropyl triethoxysilane and phenyltriethoxysilane were used as the sol–gel functional polymers to provide hydrogen bonding and π–π interactions toward the template molecules, respectively. Tetramethyl orthosilicate was used as the networking sol–gel precursor, whose primary role was to rigidly hold the template-sol–gel functional precursors complexes in the sol–gel 3D network [13,14]. The detailed MIPs synthesis process is described in Section 3.5.

Recently, molecularly imprinted polymers have aroused great interest and have been used in different fields, particularly for analytes extraction in complex samples. In the literature, just few published papers report the development of MIP-propyl paraben based and/or based on a limited selection of the above-cited molecules. Furthermore, such methods consider only environmental sample matrices. In the current study, following our research on innovative (micro)extraction procedures [15–25], different molecularly imprinted sorbent materials were studied for the extraction of most common parabens, including MPB, EPB, iPPB, PPB, iBPB, BPB, and BzPB. These sorbent phases were newly synthesized in our laboratory and herein tested in order to compare them in terms of major analytical parameters such as breakthrough volume (*VB*), retention volume (*VR*), hold-up volume (*VM*), retention factor (*k*), and theoretical plates number (*N*). Furthermore, studies on selectivity have been carried out in this work in order to obtain a full characterization for these new sorbent materials when applied on real urine samples.

#### **2. Results and Discussion**

#### *2.1. Experimental Determination of the Breakthrough Volume*

The extraction capacity in the SPE depends on many factors such as the analyte retention capacity, the loaded sample volume, the conditioning solvents type, and volumes [26].

Nowadays, many new materials are available as sorbent material for SPE. To understand the extraction mechanism of these materials and to improve the extraction efficiency, it is important to characterize them. In SPE, the most important parameters that need to be calculated are breakthrough volume (*VB*), retention volume (*VR)*, hold-up volume (*VM*), retention factor (*k*), and theoretical plates number (*N*). For the determination of breakthrough volume, a solution containing the target analytes is continually applied to the SPE cartridge containing the sorbent material; breakthrough occurs when the capacity of sorbent has been exhausted [15,26,27].

The experimental curves (concentration vs. solution volume) were fitted by means of Boltzmann's function, and the regression parameters were used in calculating the parameters (breakthrough volume, retention volume, hold-up volume, retention factor, and theoretical plates number) for each considered stationary phases loaded on SPE cartridges [15,26,27]. Figure 2 reports the Boltzmann's functions by plotting the concentration of the analyte against the aliquot volumes passed through the sorbent for each type of MIPs and for all tested analytes. These graphs represent the dependence of parabens concentrations on successive 50 mL volume loading of aqueous samples. Table 2 reports the breakthrough volume (*VB*), retention volume (*VR*), hold-up volume (*VM*), retention factor (*k*), and theoretical plates number (*N*) values calculated by using the proposed mathematical approach [15,26,27]. From the results presented in Table 2, it is evident that, as expected, the MIPs have a higher affinity for the molecule than the non-imprinted polymers (NIPs) and the polymer imprinted with BzPB has the highest breakthrough volumes, showing the highest capacity in trapping the analytes.

**Figure 2.** Boltzmann's functions determined for parabens on the different MIPs sorbent.

As reported by Bacalum et al. [27], the concentration of some compounds has different values at infinite time when passing through different MIP cartridges, and this finding could be explained by a change of interaction between analyte and the sorbent (e.g., two distinct adsorption profiles: the first herein observed, the second at higher sample volumes). Additionally, as reported by Bielica-Daszkiewicz [26] and Bacalum [27], it can be possible to observe curves where the maximum concentration used to test the sorbents not reached.

Furthermore, the curves obtained for the NIP show a shape change at very low volumes, highlighting that no retention mechanisms occur. Only for BzPB could be observed a little change, probably related to the different interactions, which could occur in relation to the chemical structure. In fact, in the presence of this analyte, larger breakthrough volume values are observed in all types of prepared MIPs, including NIPs. In this case, not only the interactions linked to the common basic structure for all parabens could be present but also there could be greater retention probably related to the second aromatic system (with related π–π interactions).




**Table 2.** *Cont*.

#### *2.2. Selectivity Study of Molecularly Imprinted Polymers*

Competitive adsorptions of methyl paraben, ethyl paraben, propyl paraben, *iso*-propyl paraben, propyl paraben, *iso*-butyl paraben, butyl paraben, and benzyl paraben were evaluated using imprinted and non-imprinted polymer. The Boltzmann's functions calculated for single paraben compared to all different molecular imprinted polymers herein tested, including NIP, were reported in Supplementary Materials, Section S.1, Figure S.1.1.

Furthermore, the selectivity that each paraben showed in its molecularly imprinted polymer compared with other imprinted polymers has been evaluated. The deviations could be related to the unselective interactions between the target analyte and the MIP phase. Table 3 shows data from the linear regression carried out to obtain comparative data of each single paraben in the various imprinted polymers, while in Supplementary Materials, the Figure S.1.2 reports the graphs for the single parabens. The values obtained for each paraben in the reference polymer have been placed on the *x*-axis, while on the *y*-axis the data relating to the single parabens in the polymers imprinted with different template are placed. This comparison was carried out in order to better highlight how the imprinted polymers are selective toward the single parabens with respect to other molecules with similar structure. By means of the obtained results, the imprinted polymers show a limited selectivity toward the single paraben, as can be observed from calculated slope values (nearest to 1 for all parabens). In particular, the imprinted polymers with PPB, iBPB, BPB, and BzPB template result more selective toward single parabens compared with MPB, EPB, and iPPB-imprinted polymers.


**Table 3.** Linear regression data.


**Table 3.** *Cont*.

However, some anomalies related to the PPB, i-BPB; BPB, and BzPB (chemical structures reported in Supplementary Materials, Figure S.1.3) have been found in MIP-PPB and MIP-BPB. MIP-PPB shows low retention capacity toward iBPB and it could be related to the chemical structure (-CH3 instead -H). The –CH3 group shows a little higher steric hindrance compared to only -H and this may partly limit the retention of this analyte for a still relatively small MIP cavity. This element can also be justified by the fact that, as shown in Table 3, the slope values are >1 for small parabens when the MIP was obtained with larger parabens as template. In addition, MIP-BPB had low retention capacity for BzPB; this is probably related to aromatic group in chemical structure of BzPB that increase the steric hindrance and not allow to a complete fit with the interaction folder.

For NIP, all values reported in Table 3 were below the unit, demonstrating that this non-imprinted polymer shows no selectivity against the considered analytes. As also highlighted in Table 2, NIP shows low breakthrough volume (mL) values in respect all the other MIPs. Furthermore, in Table 2 it was not possible to evaluate several parameters for the NIP as the Boltzmann's functions reported floating point errors.

From the analyses carried out, it appears evident that most of the selectivity of the phases is linked to the structure common to all the parabens (slope values reported in Table 3 next to unity) and that this is maximized when the same analyte is evaluated used as a different template (unitary slope). When the binding site is similar in size to the analyte, phase selectivity is not observed and interactions with the common paraben structure prevail. Similarly, if the site is larger, it is observed that structurally similar (but smaller in size) analytes are mostly retained (slope greater than 1).

For MPB the values are all close to the unit independently of the MIP. Similarly, similar behavior is observed for EPB and iPPB. Only from the PPB a distinction is made between selectivity based on the slope of the curves with values close to 1 for analytes of smaller dimensions and different from the unit (<1) for larger structures.

#### *2.3. Real Sample Analysis*

This procedure was applied to real urine sample from a healthy donor, who did not voluntarily take parabens or products containing parabens for analysis. The real sample was derived from our previous project [4], where only EPB was found in one urine real sample (as reported in Section 3.3). The urine sample, stored at −20 ◦C before analysis, was extracted following the MIP-SPE procedure reported in Section 3.7, and subsequently the eluate was injected in HPLC-PDA system. Using the MIP-EPB as sorbent in the SPE procedure, an improvement of *S*/*N* ratio of 3–5 times has been observed compared with previously obtained ratio [4], confirming the sensibility of imprinting polymers for this class of compounds. To fully compare fabric phase sorptive extraction-HPLC-PDA (FPSE-HPLC-PDA) vs. MIP-SPE-HPLC-PDA, a partial validation for MIP-SPE based procedure would be necessary, and then to compare the results by mean of statistical approach. From the point of view of the comparison between these 2 configurations, it would, however, be correct to provide for the validation of the new MIP-SPE-HPLC-PDA configuration (in order to have the best and optimized performances) and subsequently it can be applied a *t*-test.

#### *2.4. MIPs Characterization*

The evaluation of the chemical–physical characteristics of the created MIPs is reported in the Supplementary Materials, Section S.1.4. From the data obtained through Fourier transform infrared spectroscopy (FTIR), differential thermal analysis (DTA), and thermogravimetric (TGA) analysis it is possible to highlight the reproducibility of the synthesis process using different templates.

#### **3. Materials and Methods**

#### *3.1. Chemicals, Solvents, and Devices*

The International Forensic Research Institute, Department of Chemistry and Biochemistry, Florida International University (Miami, FL, USA) provided all parabens chemical standards (methyl paraben, ethyl paraben, propyl paraben, isopropyl paraben, butyl paraben, isobutyl paraben, and benzyl paraben) and all molecularly imprinted polymers.

Sodium phosphate monobasic, sodium phosphate dibasic (>99% purity grade), and phosphoric acid were obtained from Sigma-Aldrich (Milan, Italy). Acetonitrile and methanol (HPLC-grade) were purchased from Honeywell (New Jersey, USA) and were used without further purification. Deionized water (18.2 MΩ-cm at 25 ◦C) was generated by a Millipore MilliQ Plus water (Millipore Bedford Corp., Bedford, MA, USA). GraphPad Prism v.4 was used for the statistical analysis of experimental data.

#### *3.2. Stock Solution and Working Solution Preparation*

Stock solutions of chemical standards were prepared in methanol at the concentration of 1 mg/mL. The working solutions were prepared by dilution of stock solution in Milli-Q water and stored at 4 ◦C. The resulting samples were used to evaluate the enrichment factors, breakthrough volume, retention volume, hold-up volume, retention factor, and theoretical plates number.

#### *3.3. Human Urine Sample Collection, Storage, and Preparation*

Human urine sample was collected from a healthy volunteer informed about the nature of the study, who did not voluntarily take parabens or products containing parabens for analysis purposes. This urine sample is the same (a second aliquot) analyzed in the previous study [4], stored at −20 ◦C before analysis.

#### *3.4. Apparatus and Chromatographic Conditions*

The method has been described and validated by Tartaglia et al. [4]. Briefly, analyses were performed using an HPLC Thermo Fisher Scientific liquid chromatography system (Model: Spectra System P2000) coupled to a photodiode array detector (PDA) Model: Spectra System UV6000LP. Mobile phase was directly on-line degassed by using a Spectra System SCM1000 (Thermo Fisher Scientific, Waltham, MA, USA). Excalibur v.2.0 software (Thermo Fisher Scientific, Waltham, MA, USA) was used to collect and analyze data. Spherisorb C18 (15 cm × 4.6 mm, 5 μm) was used to resolve all parabens; the column was thermostated at 27 ◦C (± 1 ◦C) using a Jetstream2 Plus column oven during the analysis. The chromatographic separation was conducted in isocratic elution using phosphate buffer (28 mM, pH = 2.5) as solvent A and methanol as solvent B in volume percentages of 55 and 45, respectively. The flow rate was set at 1 mL/min. All the compounds were detected at the maximum wavelengths of 257 nm with retention time of 3.97, 6.00, 8.83, 10.43, 18.37, 19.75, and 22.33 min (for MPB, EPB, iPPB, PPB, iBPB, BPB, and BzPB, respectively).

#### *3.5. Synthesis of Novel MIPs-Parabens and NIP*

Preparation of parabens imprinted polymers involve several distinct steps: *(i)* complexation of parabens with sol–gel functional precursors; *(ii)* hydrolysis of the sol–gel cross-linking reagent; *(iii)* condensation of hydrolyzed sol–gel cross-linking reagent in presence of paraben-functional precursor complex; *(iv)* removal of paraben templates from paraben imprinted sorbents; and *(v)* synthesis of non-imprinted polymer (NIP) sorbents.

#### 3.5.1. Complexation of Parabens with Sol–Gel Functional Precursors

The complexation of individual paraben with sol–gel precursors is a spontaneous, self-assembling process directed by intermolecular interactions, e.g., hydrogen bonding, π–π interactions, and van der Waals force between the paraben template and sol–gel functional precursors. To achieve superior specificity, two sol–gel precursors phenyltriethoxysilane (PTES) and 3-aminopropyl triethoxysilane (3-APTES) were employed in the complexation process. The complexation was carried out by sequentially adding the paraben template:ethanol:PTES:3-APTES in a 50-mL centrifuge tube at a molar ratio 1:80:3:3, respectively. The mixture was vortexed vigorously after adding each ingredient. Subsequently, the solution was sonicated for 30 min. The mixture was then incubated at room temperature for 6 h so that the sol–gel functional precursors self-assembled themselves around the template molecules by hydrogen bonding and π–π interactions.

#### 3.5.2. Hydrolysis of Sol–Gel Cross-Linking Reagent

Tetramethyl orthosilicate (TMOS) was used as the sol–gel cross-linking reagent. Four methoxy functional groups connected to the central silica atom must be hydrolyzed first so that they can undergo polycondensation to build the inorganic silica network. Hydrolysis of TMOS was carried under acidic condition using trifluoroacetic acid (TFA) as the acid catalyst. The molar ratio of the template:TMOS:ethanol:TFA:water was maintained at 1:15:45:0.1:55, respectively. The mixture was prepared in a 50-mL centrifuge tube by adding individual ingredient, vortexing for 5 min and finally sonicating for 30 min. The sol solution was kept in a silicon bath at 50 ◦C for 12 h to ensure complete hydrolysis of TMOS in presence of the acid catalyst.

#### 3.5.3. Condensation of Hydrolyzed Sol–Gel Cross-Linking Reagent in Presence of the Paraben-Functional Precursors Complex

To create paraben imprinted polymers, paraben sol–gel functional precursors complexes are added to hydrolyzed sol solution in droplets under continuous stirring on a magnetic stirrer. During this process, paraben complexes randomly orient themselves within the growing sol–gel network with minimal steric hindrance and subsequently become frozen with the 3D sol–gel network. In order to complete and expedite the network formation process via condensation, the sol–gel polymer was kept in a silicon oil bath at 50 ◦C for 24 h. During this residence time in the oil bath, aging and ripening of the sol–gel network occurs, leading to a robust, highly porous silica network with trapped solvent, templates, and unreacted sol solution ingredients.

#### 3.5.4. Removal of Paraben Templates from the Imprinted Polymers

Quantitative and exhaustive removal of paraben templates from the imprinted polymer is one of the most important and challenging tasks in the creation of parabens imprinted polymers. Successful removal of the templates leave a nanocavity complimentary to the size, shape, and functionality of the template molecules and creates a highly specific synthetic receptor site, extremely affinitive towards the template molecules. To remove the templates, the paraben imprinted sol–gel polymers were first dried at 100 ◦C in a vacuum drier for 48 h. The dried MIPs are then crushed and ground in a mortar into fine particles (~50 μm) and subsequently subjected to accelerated solvent extraction at 150 ◦C and 1500 psi for 30 min using methanol as the extraction solvent. Additionally, during the analyses for breakthrough volume curves, no parabens release was observed in the conditioning steps, highlighting that the phases were completely cleaned during accelerated solvent extraction (ASE) process.

#### 3.5.5. Synthesis of Non-Imprinted Polymers

The success of molecular imprinting is often evaluated by comparing the adsorption of the template molecules on the imprinted polymer bed over a non-imprinted polymer (NIP) bed under identical conditions. Nonspecific adsorption of both the MIP and NIP can be easily estimated by exposing the known mass of the polymers into a known volume of an aqueous solution containing a known concentration of the template molecules. As such, paraben non-imprinted polymers were synthesized in parallel to the MIPs using identical process and sol solution with the only exception that no paraben templates were added in the NIP materials.

#### *3.6. Preparation of MIP–SPE Media*

A frit was placed on the bottom of an empty SPE cartridge (1 mL in total volume), which was employed as the MIP–SPE column. The SPE cartridges used in this study were in-house packed with 30 mg of parabens-MIPs (the stationary phase synthesized as previously reported in Section 3.5). Then, another frit was placed on the top of the cartridge.

#### *3.7. Molecularly Imprinted Polymer–Solid Phase Extraction Procedure*

For the breakthrough curve construction, SPE experiment was carried out as follows: 50 mL of solution at 50 μg/mL was introduced into SPE sorbent (1 mL per time), conditioned with methanol (1 mL) and water (1 mL). Each sample aliquot (1 mL) was collected in a separate vial and analyzed by HPLC-PDA to measure the concentration of analyte in each aliquot of sample. Breakthrough curves were determined as the relationship between concentration after extraction and the total volume passed through the sorbent.

For the analysis of the urine sample, the SPE cartridge loaded with molecularly imprinted polymer was conditioned by 1 mL of methanol and followed 1 mL of Milli-Q water. 1 mL of sample solution was passed through the cartridge at a flow of 1 mL/min. Subsequently, the cartridge was consecutively rinsed with 2 mL of water, and 3 mL of MeOH was used as elution solvent. Elutes (20 μL) are then injected in HPLC-PDA system.

#### **4. Conclusions**

A novel, MIP sorbent based on seven parabens template were synthesized. The new MIP-SPE method coupled with HPLC-PDA analysis was preliminarily tested for the determination of parabens in human urine, confirming that molecular imprinting technology represents a valid strategy for the synthesis of new selective extraction materials. All molecularly imprinted polymers have been characterized in terms of extraction capability, breakthrough volume, retention volume, hold-up volume, number of theoretical plates, and retention factor.

Furthermore, selectivity studies have been performed, comparing the extraction efficiency of each molecularly imprinted polymer for all the parabens tested in this study. Comparing the calculated parameters, the imprinted polymer does not show marked selectivity for every single paraben, but they provided "*class-specific*" interactions. Following this work, a polymer based on a mixture of individual MIPs for the extraction of these compounds from complex matrices could be developed. Greater efforts are therefore needed in future studies to obtain extractive materials based on imprinted polymers with increased selectivity and specificity.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/1420-3049/24/18/3334/s1, Figure S.1.1. Boltzmann's functions for single parabens., Figure S.1.2. Selectivity of each paraben., Figure S.1.3. Chemical structures of **a.** Propyl paraben; **b.** *iso*–Butyl paraben; **c.** Butyl paraben; **d.** Benzyl paraben. All the axes report the concentration values (μg/mL).

**Author Contributions:** Methodology, M.L., A.K., K.G.F., H.I.U., F.C., and G.M.M.; formal analysis, S.U., A.T., P.B., F.M., and S.G.; investigation, A.T., P.B., and F.M.; data curation, A.T.; writing—original draft preparation, A.T. and M.L.; writing—review and editing, M.L. and A.K.; supervision, M.L., A.K., K.G.F., H.I.U., C.D.O., U.d.G., and F.C, D.M.; project administration, M.L., A.K., K.G.F., F.S., H.I.U., C.D.O., U.d.G., F.C., and D.M.

**Funding:** This research received no external funding.

**Acknowledgments:** This work was supported by grant MIUR ex 60%, University of Chieti – Pescara "G. d'Annunzio", Chieti, Italy. Authors want to thank TUBITAK 2219 RESEARCH SCHOLARSHIP PROGRAM for supporting the researches of Songul Ulusoy. Authors want to thank Enrico Marcantoni at the Chemical Science Department—School of Science and Technology at the University of Camerino (Italy) for making available the FTIR and TGA instrumentations. Authors want to thank Mauro Medoro of University "G. d'Annunzio", Chieti (Italy) for his kind help in all the experiments.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


27. Bacalum, E.; Tanase, A.; David, V. Retention mechanisms applied in solid phase extraction for some polar compounds. *An. Univ. Bucar. Chim.* **2010**, *19*, 61–68.

**Sample Availability:** Samples of the compounds are available from the authors.

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
